Journal of Arid Environments 80 (2012) 10e16

Contents lists available at SciVerse ScienceDirect

Journal of Arid Environments

journal homepage: www.elsevier.com/locate/jaridenv

Effects of seed on the performance of desert legumes depend on host , stage, and density

C.W. Fox a,*, W.G. Wallin a, M.L. Bush a, M.E. Czesak b, F.J. Messina c a Department of Entomology, University of Kentucky, S225 Ag Science Center North, Lexington, KY 40546-0091, USA b Biology Department, Vassar College, Poughkeepsie, NY 12604, USA c Department of Biology, Utah State University, Logan, UT 82322-5305, USA article info abstract

Article history: Seeds of many arid habitat have a water-impermeable coat and can germinate only after being Received 23 May 2011 scarified. Bruchine seed beetles are important parasites of legume seeds in these environments, but their Received in revised form effect on germination can be unpredictable. Beetles deplete seed resources and can kill the embryo but 1 December 2011 also scarify seeds. We investigated the effects of a generalist parasite, Stator limbatus, on the germination Accepted 19 December 2011 and growth of two common legumes in the , catclaw acacia (Acacia greggii) and blue Available online 9 January 2012 paloverde ( florida). Feeding damage from a single larva greatly increased germination of paloverde but not acacia. This benefit was reduced if seeds were attacked by multiple larvae. Beetle- Keywords: Acacia damaged seeds of both hosts germinated more quickly than did control seeds. Infestation by beetles Germination reduced seedling size, though effects were greater for paloverde than for acacia. Our results demonstrate Paloverde that the effect of S. limbatus can be highly host-specific. In addition, beetle infestation may enhance or Scarification reduce seedling recruitment, depending on the availability of other scarifying agents and the number of Seed parasite larvae per seed. Such contingencies make it difficult to predict the net effect of seed beetles on efforts to Seedling growth control invasive legume hosts or establish native hosts during aridland restoration. Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction (scarification) that allows water imbibition. Seeds of many arid habitat plants have evolved in association with a variety of scari- Seed germination is influenced by many abiotic and biotic fying agents, including fire, floods, soil abrasion, and soil microor- factors, including rainfall, temperature, soil properties, leaf litter, ganisms (Bewley and Black, 1994). Germination and seedling burial depth, and vegetative cover (Bewley and Black, 1994). The recruitment may also be improved by ingestion and defecation of timing of germination can have profound effects on seedling seeds by mammals and birds (Miller, 1994a; Rohner and Ward, survival and later life-history traits in plants (Kalisz, 1986). Deter- 1999; Traveset et al., 2001). Finally, scarification can be mediated mining the factors that affect germination and early seedling by damage, which often produces holes in the seed testa growth is thus necessary to predict the establishment and regen- (Mucunguzi, 1995; Takakura, 2002). eration of plants in native habitats and for choosing appropriate Seed beetles (Coleoptera: Chrysomelidae: Bruchinae) are espe- plants to restore degraded lands (Sy et al., 2001). Many desert cially common predators and parasites of legume seeds in desert legumes play an important role in restoration ecology because they and tropical environments, and many are highly host-specific can tolerate aridity, fix atmospheric nitrogen, and provide food, (Ramirez and Traveset, 2010). Females generally lay eggs on or microhabitats, and shade for a diversity of organisms (Barnes, inside fruits, or they glue their eggs directly to seeds that are 2001; Camargo-Ricalde et al., 2004; Rohner and Ward, 1999). exposed in dehiscent pods or are found on the soil surface In arid and semi-arid environments, germination and seedling (Southgate, 1979). Larvae typically bore into the seed and complete growth are often limited by water availability (Bowers et al., 2004). development within a single seed. In some cases, seed-beetle larvae Many arid habitat legumes produce seeds with an impermeable may attack a majority of seeds in the local plant population, but coat, so that seeds can germinate only after damage to the seed coat infestation rates of some hosts can be chronically low (Miller, 1994a; Takakura, 2002). The net effects of seed beetle infestation on the germination and * Corresponding author. Tel.: þ1 859 257 7474; fax: þ1 859 323 1120. recruitment of host legumes can be unpredictable (Southgate, E-mail address: [email protected] (C.W. Fox). 1979). In some cases, the insect clearly acts as a seed predator;

0140-1963/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.jaridenv.2011.12.008 C.W. Fox et al. / Journal of Arid Environments 80 (2012) 10e16 11 larval feeding effectively kills the embryo or removes so much insect (Fox et al., 2010), we experimentally infested each host endosperm that the seed cannot germinate (Camargo-Ricalde et al., population with the same source population of S. limbatus.A 2004; El Atta, 1993; Tomaz et al., 2007). Larval feeding may also laboratory population of S. limbatus was established >300 beetles create openings for pathogenic bacteria and fungi (Chang et al., that had been collected from >20 A. greggii at the Oracle site. 2011; Cipollini and Stiles, 1991; Mucunguzi, 1995). Even if germi- Female S. limbatus oviposit directly onto host seeds inside fruits nation occurs, prior infestation of seeds may distort the develop- that have dehisced or have been damaged by other organisms, such ment of cotyledons or prevent the formation of true leaves (Hegazy as mice or other (including other seed beetles, Mitchell, and Eesa, 1991). Depletion of cotyledon reserves may slow plant 1977). Hatching larvae burrow into the seed beneath the oviposi- growth and hence reduce the probability of establishment. Beetle tion site. Because larvae cannot move between seeds, we could damage may even negatively affect non-infested seeds developing control larval density by manipulating the initial number of eggs in the same pods as infested ones, e.g., by mediating allocation of per seed. resource among seeds within the fruit (De Menezes et al., 2010). Despite these detrimental effects of seed-beetle damage, some 2.2. Treatments proportion of infested seeds will germinate successfully in most host populations (Halevy, 1974; Hoffman et al., 1989; Mack 1998; Seeds from each source population were divided among three Miller, 1994b,c). The fate of a damaged seed often depends on its treatments: control (no manipulation), beetle damage, or scarifi- size relative to that of its parasite, as well as the number of larvae cation. We scarified seeds by removing about 5% of the seed coat per seed (Fox et al., 2010). If beetle damage acts as a scarifying with sandpaper, which should promote imbibition of water agent, it can have a large, positive effect on the frequency of without affecting the underlying cotyledons. This technique has germination (Arévalo et al., 2010; Mucunguzi, 1995; Nakai et al., been widely used for physical scarification of desert legume seeds 2011). In one study, beetle infestation was considered a prerequi- (e.g., Patane and Gresta, 2006). To produce beetle-damaged seeds, site to successful germination (Takakura, 2002). Because their mated pairs of beetles were individually confined with five pre- effects on host performance can be so variable, bruchine beetles weighed seeds in a 30-mm Petri dish. Providing only five seeds have been implicated in aiding the spread of undesirable, invasive per dish caused females to lay multiple eggs per seed. Females were legumes (Arévalo et al., 2010) and, conversely, have been proposed allowed to lay eggs for w72 h and discarded. Eggs were scraped as biological control agents for such invasive hosts (van Klinken and from seeds to create densities ranging from one to five eggs per Flack, 2008). seed. Larvae within seeds were reared to adult emergence in This study examines the effects of seed beetle infestation and a laboratory growth chamber at 30 C. Because nearly all eggs hatch damage on two desert legumes: catclaw acacia, Acacia greggii Gray successfully (>95% of eggs hatch; Fox et al., 2007), the number of (: Mimosoideae), and blue paloverde, Parkinsonia florida eggs on a seed is a good estimate of larval density. Seeds from the (Bentham ex A. Gray) S. Watson (Fabaceae: ). Both three treatments were randomly interspersed in the chamber to species are common in the southwestern United States and ensure that all seeds experienced similar environmental northwestern Mexico. Germination of acacia and paloverde seeds is conditions. likely to be enhanced by scarification, which in may be After adult emergence from seeds had been completed, we accomplished by flash floods in gravelly soils, underground removed beetle frass from seeds by tapping the seed and inserting weathering, or ingestion by mammals (Longland et al., 2001; Or a small brush into beetle exit holes. Seeds were then reweighed to and Ward, 2003). Seeds of A. greggii and P. florida are also estimate the amount of mass removed by beetle infestation. frequently attacked by a generalist seed beetle, Stator limbatus Horn Because larval survival can be fairly low in seeds of blue paloverde (Fox et al., 1995, 1997; Siemens et al., 1992). This insect has a rela- (Fox et al., 1995; Fox and Mousseau, 1996), we carefully noted the tively wide geographic distribution and a broad host range total number of adults emerging from each seed. To remove biases (Stillwell et al., 2007, and references therein), but little is known caused by the presence of dead larvae inside seeds, we estimated about this beetle’s impact on host performance. To determine the beetle effects on plant performance using only those seeds from effect of S. limbatus on seedling recruitment in A. greggii and which all beetles emerged successfully as adults. P. florida, we experimentally manipulated densities of S. limbatus in seeds and measured effects on both germination and seedling 2.3. Germination and seedling growth growth. Control, infested, and mechanically scarified seeds were sown in 2. Materials and methods pots in a University of Kentucky greenhouse furnished with supplemental lighting and maintained at 29 1 C (day) and 2.1. Source populations of seeds and insects 27 1 C (night) and a 15:9 L:D photocycle. Seeds were sown in 13 blocks over a year, with both species and treatments represented in Ò Seeds were obtained from three sites in central Arizona: two each block. Each pot was filled with loosely packed ProMix and populations of catclaw acacia and one population of blue paloverde. received a single seed 2.5 cm below the soil surface. Pots were Seeds of the Oracle population of A. greggii were collected from watered daily. Successful germination was noted when seedling mature pods on >20 trees along Hwy 77 and adjacent roads in tissue became visible above the soil surface. Seeds were classified Oracle, Pinal County (32.62 N 110.78 W). Seeds of the Phoenix as having failed to germinate if three months had elapsed with no population of A. greggii were similarly collected from >20 trees in visible tissue above the soil surface. This protocol would not detect the Cave Creek area north of Phoenix, Maricopa County (w32.4 N instances in which seeds did germinate but seedlings died before 112.0 W). Seeds of blue paloverde were also collected from >20 emerging from the soil. However, later inspection of pots with no trees in Maricopa County (33.79 N112.12 W). Based on a sample visible seedlings indicated that such ‘cryptic’ germination was rare of >900 seeds from each population, average seed mass (SD) was and unlikely to influence any of the treatment effects. Fifteen days 195 70 mg for Oracle acacia, 205 72 mg for Phoenix acacia, and after germination, we measured the height of each seedling, care- 199 50 mg for blue paloverde. fully washed it to remove clinging soil, and dried it to constant Because the effects of insect damage on seed germination and weight at 60 C. We used an electronic balance to measure above- plant growth may vary depending on the source population of the and belowground dry mass of each seedling. 12 C.W. Fox et al. / Journal of Arid Environments 80 (2012) 10e16

We planted 1913 seeds (560 Oracle acacia, 547 Phoenix acacia, paloverde and acacia), but the mass removed by a single larva did and 806 paloverde), of which 837 (171, 322, and 344) plants not differ between the two source populations of acacia (t372 ¼ 1.25, germinated and were grown. P ¼ 0.21). Increasing beetle density predictably increased the total mass lost for seeds of both acacia and paloverde (Table 1). However, 2.4. Analyses the amount of seed mass lost per larva was not affected by larval density in Oracle acacia seeds (F1,413 ¼ 0.26, P ¼ 0.6) or in paloverde ¼ ¼ Logistic regression was used to compare the frequency of seeds (F1,102 2.08, P 0.15). For acacia seeds from the Phoenix germination among treatments. All other traits were subjected to population, the amount of mass lost per larva decreased slightly but fi ¼ ¼ analyses of variance to test for heterogeneity among classes (Proc signi cantly with increasing larval density (F1,420 4.0, P 0.05). MIXED, SAS Institute, Cary, North Carolina). Linear contrasts between particular treatment combinations were performed using 3.2. Frequency and timing of germination the ESTIMATE statement. Initial seed mass and mass lost to beetle feeding were used as covariates in some analyses. We present F The proportion of control seeds that germinated varied among c2 ¼ < statistics for the overall ANOVA and ANCOVA results and a t statistic the three host populations (logistic regression; 2 117, P 0.001). for linear contrasts. For all parametric analyses, “block” was Undamaged paloverde seeds germinated at a much lower included as a random effect. This effect was almost always signifi- frequency than seeds from either acacia population (Fig. 1A; c2 > < cant, but was included only to control for among-block variation, 1 48, P 0.001 for both comparisons). In addition, control acacia and is not discussed further below. We present least-squares means seeds from the Phoenix population germinated at a higher c2 ¼ to remove block effects on plant traits. These were estimated with frequency than those from the Oracle population ( 1 15.6, < the SAS PROC MIXED procedure. Estimates of least-squares means P 0.001) (Fig. 1A). The frequency of germination of acacia seeds fi depend on the particular statistical model, and may differ slightly was not signi cantly affected by infestation by a single larva of c2 < > between analyses that included different treatment combinations. S. limbatus ( 1 0.46, P 0.49), but beetle infestation dramatically Because we weighed and sowed only seeds from which all increased the proportion of paloverde seeds that germinated c2 ¼ < larvae emerged as adults, we obtained relatively low sample sizes (Fig. 1A; 1 142, P 0.001). fi fi of seeds infested with four or five larvae. Sample sizes were Mechanical scari cation signi cantly improved germination c2 > < adequate for analyses of germination frequency (for which both relative to control seeds (Fig. 1A; 1 5, P 0.03 for each of the germinated and non-germinated seeds are included). However, three analyses), but it had an especially large effect on germination fi germination frequencies were low for these highly infested seeds of paloverde, which rose from 5% to 95% (Fig. 1A). Scari ed seeds (see Results), resulting in very low sample sizes for plant growth were also more likely to germinate than singly infested seeds in all c2 > < traits at these higher densities. We thus included all five larval host populations (Fig. 1A; 1 4.2, P 0.05). Thus, rankings of fi > ¼ densities in the analysis of germination frequency, but excluded germination frequency were scari ed control infested for fi >> >> larval densities greater than three beetles/seed in the analyses of acacia, whereas they were scari ed infested control for germination time and seedling growth.

3. Results A 1.0 Control seeds 3.1. Loss of mass in infested seeds Scarified seeds Stator-damaged seeds 0.8 Infestation by a single larva reduced seed mass by about 6% in acacia seeds (both populations) and about 4% in paloverde seeds 0.6 (Table 1). This difference between legume species was highly significant (t372 ¼ 11.12, P < 0.001 for the contrast between 0.4 0.2 Proportion germinated 0.0 Table 1 0.50 Acacia Acacia Paloverde Mean mass lost (SE) by catclaw acacia and blue paloverde seeds after infestation by Oracle Phoenix varying densities of the seed beetle Stator limbatus.

Host Larval No. of Total mass Mass lost Proportion of B 0.7 Oracle acacia population density seeds lost (mg) per larva initial mass Phoenix acacia (mg) lost 0.6 paloverde Catclaw 1 156 11.0 0.2 11.0 0.2 0.062 0.001 0.5 acacia (Oracle) 2 109 22.1 0.4 11.0 0.2 0.115 0.003 3 72 32.7 0.5 10.7 0.2 0.165 0.005 0.4 4 52 43.8 1.0 11.0 0.2 0.203 0.006 0.3 5 28 55.5 1.4 11.1 0.3 0.229 0.008 0.2 Catclaw 1 161 11.3 0.2 11.3 0.2 0.058 0.001 acacia (Phoenix) 2 119 22.1 0.3 11.0 0.2 0.112 0.002 0.1 Proportion germinated 3 82 32.6 0.4 10.9 0.1 0.153 0.004 0.0 4 39 41.9 0.8 10.5 0.2 0.190 0.005 012345 5 23 52.0 1.5 10.4 0.3 0.217 0.011 Number of beetle larvae per seed Blue paloverde 1 58 7.6 0.2 7.6 0.2 0.043 0.002 Fig. 1. A. Frequency of germination (mean SE) of control seeds, mechanically scari- 2 23 16.0 0.6 8.0 0.3 0.080 0.003 fied seeds, and seeds previously infested by a single Stator limbatus larva. Seeds were 3 18 23.3 1.0 7.8 0.3 0.120 0.007 from two populations of catclaw acacia and one population of blue paloverde. B. Mean initial seed mass (SE) was 195 2, 205 2, 199 1 mg for Oracle acacia, Germination frequencies of seeds previously infested with varying densities of Phoenix acacia, and blue paloverde, respectively (N > 900 seeds per population). S. limbatus. Some error bars are not visible because they are smaller than the points. C.W. Fox et al. / Journal of Arid Environments 80 (2012) 10e16 13 paloverde (Fig.1A). It was also evident that Phoenix acacia seeds are 3.3. Seedling growth consistently more likely to germinate than are Oracle acacia seeds subjected to the same treatment (Fig. 1A). Not surprisingly, larger seeds produced larger seedlings. In an For all host populations, germination success generally analysis of covariance, there was a positive relationship between decreased with an increasing number of larvae per seed (Fig. 1B; seed mass and the height and mass (above- and belowground) of 2 c1 > 7.2, P < 0.008 for each host). However, the severity of this control seedlings at 15 days after germination (F > 5.4, P < 0.001 for decline depended on host population, i.e., there was a significant all traits). However, the effect of seed size differed between acacia 2 host population larval density interaction (c2 ¼ 20.4, P < 0.001). and paloverde; the effect was large and highly significant for both Despite gaining a very large benefit in germination frequency from acacia populations (P < 0.001 for all traits) and either non- infestation by a single larva, paloverde seeds then suffered the significant (for height and belowground biomass) or only margin- steepest decline in germination frequency with an increasing ally significant (P ¼ 0.03 for aboveground biomass) for paloverde number of larvae per seed (Fig. 1B). Seeds from both species seedlings. germinated at a very low frequency if they bore five larvae per seed The effect of infestation by one beetle larva depended on the (Fig. 1B). type of host; for all traits, there was a significant interaction effect Among control seeds, acacia took much longer to germinate between host and the presence or absence of a single larva (Fig. 3). than did paloverde (t203 ¼ 4.8, P < 0.001; Fig. 2A). Although infes- There were no differences in height among seedlings from control, tation by a single larva had no effect on germination frequency in scarified or singly infested seeds for either acacia population, acacia (Fig. 1A), it reduced the time required for germination by (F2,441 2.19, P > 0.33 for each population), but the three treatment >65% in all three host populations (Fig. 2A; t768 ¼ 4.17, P < 0.001 for groups (control, scarified, and singly infested seeds) did vary in the comparison of control versus one-larva seeds). Because above- and belowground mass (F2,439 > 6.1, P < 0.003 for each mechanical scarification also greatly reduced germination time in trait); infestation by a single beetle larva reduced biomass all populations (t768 ¼ 4.72, P < 0.001), there was no difference in compared to seeds not damaged by a beetle larvae (t439 > 2.1, the time required for germination between scarified and singly P < 0.04). Paloverde seedlings germinating from singly infested infested seeds (Fig. 2A; t768 ¼ 0.18, P ¼ 0.86). Seed mass had no seeds were significantly smaller for all three plant traits than effect on the time to germination of beetle damaged seeds for any seedlings germinating from seeds not infested by a beetle (Fig. 3; host (F < 1.0, P > 0.33), and there was no relationship between the P < 0.001 for each trait), but there were no significant differences proportion of mass lost to feeding by a single larva and the time needed for germination (F < 1.7, P > 0.20 for each host population). Although infestation by at least one larvae substantially reduced 20 the time until germination, the number of larvae per seed (from 1 A to 3) had no effect on the time needed for germination (Fig. 2B; 18 Control seeds ¼ ¼ Scarified seeds F1,240 0.91, P 0.34). 16 Stator-damaged seeds 14 12

Height (cm) 10 A Control seeds 8 35 Scarified seeds Stator-damaged seeds 6 30 25 300 B 20 15 200 10

Days to germination 5 0 100 Acacia Acacia Paloverde3.40 Oracle Phoenix

Aboveground biomass (mg) 0 B 35 Oracle acacia 70 C 30 Phoenix acacia paloverde 25 60 20 50 15 10 40

Days to germination 5 30 0 0123 Belowground biomass (mg) 20 Number of beetle larvae per seed 0.50 Acacia Acacia Paloverde Oracle Phoenix Fig. 2. A. Time required for germination (mean SE) of control seeds, mechanically scarified seeds, and seeds previously infested by a single Stator limbatus larva. B. Fig. 3. AeC. Height and dry mass (mean SE) of seedlings derived from control seeds, Germination times of seeds previously infested with varying densities of S. limbatus. mechanically scarified seeds, and seeds previously infested by a single Stator limbatus Some error bars are not visible because they are smaller than the points. larva. Some error bars are not visible because they are smaller than the points. 14 C.W. Fox et al. / Journal of Arid Environments 80 (2012) 10e16 between seedlings from control and scarified seeds (P > 0.43 for consumption had a significant effect on all three measures of each trait). There was no significant difference between the two seedling growth (F > 7.0, P < 0.007 for all traits). Interestingly, acacia populations for any of the plant traits (Fig. 3; P > 0.11 for all though, the effect of beetle density remained significant after three traits). For all seedling traits in acacia, the effect of beetle accounting for the amount of seed mass lost (using analysis of damage also depended on initial seed mass, i.e., there was covariance) for above- and belowground biomass (F > 4.4, P < 0.03), a significant interaction between the effects of seed mass and but not height (F1,91 ¼1.33, P ¼ 0.25) of acacia, and for height and infestation by a single larva (F > 12.0, P < 0.001 for all traits). As aboveground biomass (F > 5.2, P < 0.03), but not belowground expected, smaller seeds were much more negatively impacted by biomass (F1,130 ¼ 1.54, P ¼ 0.22), of paloverde. This indicates that beetle damage. In contrast, the effect of beetle damage also did not increasing beetle density reduced the size of seedlings beyond that depend on initial seed mass for paloverde seedlings (F < 0.7, accounted for by the increased amount of seed tissue lost to beetles P > 0.4). in highly infested seeds. Increasing the number of larvae per seed consistently reduced seedling height and biomass (Fig. 4; F > 4.2, P < 0.04 for all traits in 4. Discussion an ANCOVA that included only beetle damaged seeds). There was no evidence that the effect of density differed among the three host Taken together, our results highlight the complex and mixed populations, as there was no interaction between the effects of effects of seed-beetle infestation on the performance of their native larval density and host for any seedling trait (Fig. 4; F < 2.7, P > 0.07 legume hosts. Because the body size of S. limbatus is small relative for each trait). As we observed for control seeds, seed mass was to the size of an acacia or paloverde seed, seeds supporting the positively correlated to the size of beetle-damaged seedlings for complete development of one larva lost only a small fraction of acacia (F > 11.0, P < 0.001 for all three traits) but not for paloverde their initial mass. Consequently, damage to embryonic tissue was (F < 1.0, P > 0.5). However, there was no evidence that seed mass less, and there was no decrease in the frequency of germination of mediated the effects of beetle density on seedling size (there was singly infested seeds relative to control seeds for either species. On no seed mass-x-beetle density interaction for any trait of either the contrary, feeding by a single larva greatly improved the species; F < 1.0, P > 0.47). The proportion of seed mass lost to beetle germination success of blue paloverde, which relies on scarification much more than does catclaw acacia. However, the effect of beetle infestation was density-dependent; infestation by multiple larvae reduced the germination frequency of both species, but especially of paloverde seeds. Since acacia seeds, unlike paloverde seeds, did A not benefit from infestation by a single larva, the effect of S. limba- Oracle acacia 20 Phoenix acacia tus on the germination success of acacia would be neutral (in singly paloverde infested seeds) or detrimental (in multiply infested seeds) in rela- 16 tion to germination of control seeds. 12 Many legumes have seed coats that are largely impermeable to water, limiting their ability to imbibe water, an essential first event 8 for germination (Takakura, 2002). Such species often require, or at

Height (cm) least benefit from, scarification of their seed coat (Clemens et al., 4 1977). Our study demonstrates that beetle damage may be an fi 0 important scari cation mechanism in nature, but also that the 0123 effect of beetle damage on host germination must be considered in B the context of other potential scarifying agents. For both species fi 300 examined here, mechanical scari cation improved germination relative to both control and beetle damaged seeds. This was espe- cially true for paloverde, which is very dependent on scarification 200 before it can germinate. For paloverde seeds, infestation by a single larva was beneficial relative to control seeds (beetle damage dramatically increased germination), but was costly relative to 100 mechanically scarification (beetle damage dramatically reduced germination). Apparently, even one larva removes enough seed tissue to reduce the benefit of creating an opening in the seed coat.

Aboveground biomass (mg) 0 0123 Optimal germination of both acacia and paloverde seeds would C occur via physical disruption of the seed coat without insect colo- 80 nization. Nevertheless, seed beetles could be an important scarifi- cation agent in natural populations of blue paloverde, since 60 candidate abiotic processes, such as scraping of seeds during flash floods or weathering of seeds under the soil surface (Bewley and 40 Black, 1994), may not be as frequent or effective as insect- or rodent-mediated scarification (cf., Mitchell, 1977; Takakura, 2002). 20 More research is needed with respect to longer-term effects of either mechanical or beetle-mediated scarification. It is possible,

Belowground biomass (mg) 0 for example, that either type of damage to the seed coat increases 0123 desiccation and reduces seed longevity. Number of beetle larvae per seed In some desert species, seed dormancy is broken when germi- nation inhibitors are leached from the seed coat after sufficient Fig. 4. AeC. Height and dry mass (mean SE) of seedlings derived from seeds previously infested by varying densities of Stator limbatus. Some error bars are not rainfall, or upon dilution of salt in the soil, ensuring that germi- visible because they are smaller than the points. nation occurs when the soil moisture level can support seedling C.W. Fox et al. / Journal of Arid Environments 80 (2012) 10e16 15 growth (Bewley and Black, 1994). Commonly, early germinating lower survival in harsh habitats (Sher et al., 2004). In nature, beetle seeds (e.g., those germinating early in the rainy season) have damage would also reduce seedling survival if it altered the a fitness advantage over later germinating seeds because they may seasonal timing of germination in a maladaptive way; e.g., if insect- experience a longer growing season (Gómez, 2004), although there caused scarification reduces the total amount of soil moisture are many exceptions (Kalisz, 1986, and references therein; Cheplick, needed for germination, seeds may germinate when average rain- 1996). Delayed germination may in fact be beneficial if it increases fall is suboptimal for seedling growth and persistence (Elberse and the total variance in germination times and makes it more likely Bremen, 1990). that at least some individuals germinate under temporarily favor- Our study also does not consider other ecological interactions able conditions. For the desert species used in our study, infestation that may have profound effects on germination and seedling of seeds by beetles substantially shortened the time required for growth in nature (Rodriguez-Pérez et al., 2011). For example, seeds germination post-watering (by w10 days for paloverde and >15 cached in rodent dens provide the major source of recruitment for days for acacia). However, this benefit of beetle damage was many desert (McAuliffe, 1990), but damaged seeds may generally less than the benefit of mechanical scarification, indi- more likely be eaten than cached, as has been demonstrated for cating again that the net effect of insect damage depends on the squirrel predation on acorns (Steele et al., 1996). Alternatively, likelihood of mechanical scarification. some rodents preferentially harvest undamaged seeds, such that Despite the observed positive effects on germination frequency low levels of insect damage may protect seeds from vertebrate and timing, we must be cautious before concluding that low levels predators (Miller, 1994b). Beetle damage may also reduce the of beetle attack are generally beneficial. After germination, beetle beneficial effects of other scarifying agents. For example, coyotes, damage to seeds has lingering, but species-specific, negative effects woodrats, deer and agricultural ruminants consume large numbers on seedlings. The negative effect of infestation by a single S. lim- of desert seeds, many of which pass through the gut scarified and batus larva on 15-day-old seedlings was small for acacia seedlings ready to germinate (Kneuper et al., 2003). However, insect- but substantial for paloverde seedlings. It is not clear why this damaged seeds may be more likely to be destroyed by mamma- between-species difference occurred, since seeds of the two species lian gut enzymes (Miller, 1993). We also do not know whether are similar in size, and the amount of mass lost by a singly infested improved germination rates of singly infested seeds in this study seeds was actually lower for paloverde than for acacia. The con- would be somewhat lower in nature, where insect feeding may trasting responses of acacia and paloverde may reflect fundamental facilitate fungal attack (e.g., Chang et al., 2011). differences in how seed reserves are used in each species. In some plants, part of the cotyledon may serve to buffer the negative impact of pre-dispersal seed predation, whereas any loss of coty- 5. Broader implications ledon tissue in other plants greatly reduces seedling mass (Bonal et al., 2007; Mack, 1998). Control paloverde seeds produced larger Our study demonstrates that the overall effect of S. limbatus on seedlings than did control acacia seeds (Fig. 3), which may indicate the germination and growth of its legume host can range from that paloverde uses proportionally more reserves for seedling highly negative to positive, depending on host species, plant stage, growth, and any loss of cotyledon tissue has negative conse- and larval density. At one extreme, attack by a single beetle quences. For paloverde, the negative effects of beetle damage on improves the frequency and timing of germination of paloverde seedling performance would partially offset the benefits of seeds, and thus can have very positive effects on plant fitness in the increased germination frequency and decreased time to germina- absence of other seed scarifying agents. At the other extreme, tion, but this is not the case for acacia seeds. moderate beetle densities (within ranges commonly observed in In Arizona, it is typical for acacia and paloverde seeds to be nature) reduce germination frequency (probably by killing the attacked by multiple species of seed beetles in the Sonoran desert, embryo), and reduce growth of seedlings, substantially reducing and individual seeds often support multiple beetle larvae (C. Fox, plant fitness. It should not be surprising therefore that previous pers. obs). For example, it is common for nearly all paloverde seeds studies have reported highly beneficial, moderate, and highly to be attacked by either S. limbatus or another bruchine beetle detrimental effects of seed beetles on the fates of host seeds ( amicus), with more than half of all seeds producing at (Arévalo et al., 2010; De Menezes et al., 2010; Ernst et al., 1989; Or least one emerging adult beetle (Mitchell, 1977). Of seeds that are and Ward, 2003; Takakura, 2002; Tomaz et al., 2007). Even attacked by beetles, most receive multiple eggs. For example, an congeneric Acacia species in Africa savannas showed opposite average blue paloverde seed is attacked by >3 Mimosestes eggs responses to attack by a single beetle species (Mucunguzi, 1995). (Siemens et al., 1992), and most seeds attacked by S. limbatus Such complexities in the effects of beetle damage on plant receive more than one egg (Amarillo-Suarez, 2010; unpubl. data). performance make it difficult to quantify the role of a seed parasite We thus included variable beetle density in our study, and the in limiting or advancing the spread of invasive plants (Arévalo et al., results are generally clear and consistent for both species e the 2010; Briano et al., 2002; van Klinken and Flack, 2008), or in frequency of seed germination and seedling growth both decline decreasing population sizes of endangered plants (Hegazy and quite substantially with increasing numbers of beetles, such that Eesa, 1991). Host-specific and density-dependent effects also any benefit of beetle damage caused by scarification of the seed make it difficult to predict how seed parasites will slow the re- coat (e.g., in paloverde) is eliminated. In other seed/seed-parasite establishment of native legumes in degraded or denuded habitats associations, the frequency and consequences of multiply infested (Camargo-Ricalde et al., 2004; Khurana and Singh, 2001). To obtain seeds can be a complex function of the overall abundance of the an even more realistic assessment of the effect of a seed parasite on seed crop as well as the average sizes of individual seeds (Bonal its host, one should also examine potential multi-species interac- et al., 2007). tions involving seed beetles and their hosts (Brancalion et al., 2011; Our experiments probably underestimated the negative effects Mitchell, 1977; Miller, 1994aec). For example, if seed beetle larvae of beetle damage on seedling performance in nature. Experimental are attacked by parasitoid wasps, they may accomplish scarification plants were provided ample water at regular intervals, and were but not consume enough tissue to kill seeds (Nakai et al., 2011). reared in ProMix in the absence of competitors. Water availability is Despite these caveats, experimental manipulations such as those in a major determinant of seedling survival and growth in arid habitat this study can begin to disentangle the complex interactions of plant species, and even a modest decrease in initial plant size may seed-feeders and plants in natural communities. 16 C.W. Fox et al. / Journal of Arid Environments 80 (2012) 10e16

Acknowledgments Hegazy, A.K., Eesa, N.M., 1991. On the ecology, insect seed-predation, and conser- vation of a rare and endemic plant-species: Ebenus armitagei (Leguminosae). Conservation Biology 5, 317e324. We thank Kim Carrico, R. Nathan Correll, E. Courtney Elkins, J. Hoffman, M.T., Cowling, R.M., Douie, C., Pierce, S.M., 1989. Seed predation and Matt Fulkerson, Laura Hettinger, Lisa Hitchcock (Boock), Oriaku germination of Acacia erioloba in the Kuiseb river valley, Namib desert. South e Njoku, and Jennifer Smith for laboratory and greenhouse assis- African Journal of Botany 55, 103 106. Kalisz, S., 1986. Variable selection on the timing of germination in Collinsia verna tance. This research was supported by the University of Kentucky (Scrophulariaceae). Evolution 40, 479e491. Agricultural Experiment Station (to CWF) and the Utah Agricultural Khurana, E., Singh, J.S., 2001. Ecology of seed and seedling growth for conservation Experiment Station (to FJM; UAES paper no. 8301). and restoration of tropical dry forest: a review. Environmental Conservation 28, 39e52. Kneuper, C.L., Scott, C.B., Pinchak, W.E., 2003. Consumption and dispersion of mesquite seeds by ruminants. Journal of Range Management 56, 255e259. References Longland, W.S., Jenkins, S.H., Vander Wall, S.B., Veech, J.A., Pyare, S., 2001. Seedling recruitment in Oryzopsis hymenoides: are desert granivores mutualists or Amarillo-Suarez, A.R., 2010. Top-down, bottom-up, and horizontal mortality vari- predators? Ecology 82, 3131e3148. ation in a generalist seed beetle. Revista Colombiana de Entomología 36, Mack, A.L., 1998. An advantage of large seed size: tolerating rather than succumbing 269e276. to seed predators. Biotropica 30, 604e608. Arévalo, J.R., Afonso, L., Naranjo, A., Salas, M., 2010. Invasion of the Gran Canaria McAuliffe, J.R., 1990. Paloverdes, pocket mice, and bruchid beetles: interrelation- ravines ecosystems (Canary Islands) by the exotic species Acacia farnesiana. ships of seeds, dispersers, and seed predators. Southwestern Naturalist 35, Plant Ecology 206, 185e193. 329e337. Barnes, M.E., 2001. Seed predation, germination and seedling establishment of Miller, M.F., 1993. Is it advantageous for Acacia seeds to be eaten by ungulates? Acacia erioloba in northern Botswana. Journal of Arid Environments 49, Oikos 66, 364e368. 541e554. Miller, M.F., 1994a. Large African herbivores, bruchid beetles and their interactions Bewley, J.D., Black, M., 1994. Seeds: Physiology of Development and Germination, with Acacia seeds. Oecologia 97, 265e270. second ed. Plenum Press, New York, NY, USA. Miller, M.F., 1994b. The costs and bene fits of Acacia seed consumption by ungulates. Bonal, R., Muñoz, A., Díaz, M., 2007. Satiation of predispersal seed predators: the Oikos 71, 181e187. importance of considering both plant and seed levels. Evolutionary Ecology 21, Miller, M.F., 1994c. The fate of mature African Acacia pods and seeds during their 367e380. passage from the to the soil. Journal of Tropical Biology 10, 183e196. Bowers, J.E., Turner, R.M., Burgess, T.L., 2004. Temporal and spatial patterns in Mitchell, R., 1977. Bruchid beetles and seed packaging by palo verde. Ecology 58, emergence and early survival of perennial plants in the Sonoran Desert. Plant 644e651. Ecology 172, 107e119. Mucunguzi, P., 1995. Effects of bruchid beetles on germination and establishment of Brancalion, P.H.S., Rodrigues, R.R., Novembre, A.D.L.C., Gomez, J.M., 2011. Are we Acacia species. African Journal of Ecology 33, 64e70. misinterpreting seed predation in palms? Biotropica 43, 12e14. Nakai, Z., Kondo, T., Akimoto, S., 2011. Parasitoid attack of the seed-feeding beetle Briano, J.A., Cordo, H.A., Deloach, C.J., 2002. Biology and field observations of Pen- Bruchus loti enhances the germination success of Lathyrus japonicus seeds. thobruchus germaini (Coleoptera: Bruchidae), a biological control agent for ArthropodePlant Interactions 5, 227e234. Parkinsonia aculeata (Caesalpiniaceae). Biological Control 24, 292e299. Or, K., Ward, D., 2003. Three-way interactions between Acacia, large mammalian Camargo-Ricalde, S.L., Dhillion, S.S., Garcia-Garcia, V., 2004. Phenology, and seed herbivores and bruchid beetles e a review. African Journal of Ecology 41, production and germination of seven endemic Mimosa species (Fabaceae: 257e265. Mimosoideae) of the Tehuacan-Cuicatlan Valley, Mexico. Journal of Arid Envi- Patane, C., Gresta, F., 2006. Germination of Astragalus hamosus and Medicago orbi- ronments 58, 423e437. cularis as affected by seed-coat dormancy breaking techniques. Journal of Arid Chang, S.M., Gonzalez, E., Pardini, E., Hamrick, J.L., 2011. Encounters of old foes on Environments 67, 165e173. a new battle ground for an invasive tree, Albizia julibrissin Durazz (Fabaceae). Ramirez, N., Traveset, A., 2010. Predispersal seed-predation by insects in the Ven- Biological Invasions 13, 1043e1053. ezuelan Central Plain: overall patterns and traits that influence its biology and Cheplick, G.P., 1996. Do seed germination patterns in cleistogamous annual grasses taxonomic groups. Perspectives in Plant Ecology Evolution and Systematics 12, reduce the risk of sibling competition? Journal of Ecology 84, 247e255. 193e209. Cipollini, M.L., Stiles, E.W., 1991. Seed predation by the Acanthoscelides Rodriguez-Pérez, J., Wiegand, K., Ward, D., 2011. Interaction between ungulates and obtectus on Phaseolus species: consequences for seed size, early growth and bruchid beetles and its effect on Acacia trees: modeling the costs and benefits of reproduction. Oikos 60, 205e214. seed dispersal to plant demography. Oecologia 167, 97e105. Clemens, J., Jones, P.G., Gilbert, N.H., 1977. Effect of seed treatments on germination Rohner, C., Ward, D., 1999. Large mammalian herbivores and the conservation of of Acacia. Australian Journal of Botany 25, 269e276. arid Acacia stands in the Middle East. Conservation Biology 13, 1162e1171. De Menezes, L.C.C.R., Klein, J., Kestring, D., Rossi, M.N., 2010. Bottom-up and top- Sher, A.A., Goldberg, D.E., Novoplansky, A., 2004. The effect of mean and variance in down effects in a pre-dispersal seed predation system: are non-predated resource supply on survival of annuals from Mediterranean and desert envi- seeds damaged? Basic and Applied Ecology 11, 126e134. ronments. Oecologia 141, 353e362. El Atta, H.A., 1993. The effect of Caryedon serratus Olivier (Col.: Bruchidae) on Siemens, D.H., Johnson, C.D., Ribado, K.J., 1992. Alternative seed defense mecha- viability and germination of seeds of Acacia nilotica (L.Will) in the Sudan. Forest nisms in congeneric plants. Ecology 73, 2152e2166. Ecology and Management 57, 169e177. Southgate, B.J., 1979. Biology of Bruchidae. Annual Review of Entomology 24, Elberse, W.T., Bremen, H., 1990. Germination and establishment of Sahelian ran- 449e473. geland species. II. Effects of water availability. Oecologia 85, 32e40. Steele, M.A., Hadj-Chikh, L.Z., Hazeltine, J., 1996. Caching and feeding decisions by Ernst, W.H.O., Tolsma, D.J., Decelle, J.E., 1989. Predation of seeds of Acacia tortilis by Sciurus carolinensis: responses to weevil-infested acorns. Journal of insects. Oikos 54, 294e300. Mammalogy 77, 305e314. Fox, C.W., Mousseau, T.A., 1996. Larval host plant affects the fitness consequences of Stillwell, R.C., Morse, G.E., Fox, C.W., 2007. Geographic variation in body size and egg size in the seed beetle Stator limbatus. Oecologia 107, 541e548. sexual size dimorphism of a seed-feeding beetle. The American Naturalist 170, Fox, C.W., Bush, M.L., Messina, F.J., 2010. Biotypes of the seed beetle Callosobruchus 358e369. maculatus have differing effects on the germination and growth of their legume Sy, A., Grouzis, M., Danthu, P., 2001. Seed germination of seven Sahelian legume hosts. Agricultural and Forest Entomology 12, 353e362. species. Journal of Arid Environments 49, 875e882. Fox, C.W., Scheibly, K.L., Smith, B.P., Wallin, W.G., 2007. Inbreeding depression in Takakura, K., 2002. The specialist seed predator Bruchidius dorsalis (Coleoptera: two seed-feeding beetles, Callosobruchus maculatus and Stator limbatus (Cole- Bruchidae) plays a crucial role in the seed germination of its host plant, Gleditsia optera: Chrysomelidae). Bulletin of Entomological Research 97, 49e54. japonica (Leguminosae). Functional Ecology 16, 252e257. Fox, C.W., Waddell, K.J., Mousseau, T.A., 1995. Parental host plant affects offspring Tomaz, C.A., Kestring, D., Rossi, M.N., 2007. Effects of the seed predator Acantho- life histories in a seed beetle. Ecology 76, 402e411. scelides schrankiae on viability of its host plant Mimosa bimucronata. Biological Fox, C.W., Waddell, K.J., des Lauriers, J., Mousseau, T.A., 1997. Seed beetle survi- Research 40, 281e290. vorship, growth and egg size plasticity in a paloverde hybrid zone. Ecological Traveset, A., Riera, N., Mas, R.E., 2001. Passage through bird guts causes interspecific Entomology 22, 416e424. differences in seed germination characteristics. Functional Ecology 15, Gómez, J.M., 2004. Bigger is not always better: conflicting selective pressures on 669e675. seed size in Quercus ilex. Evolution 58, 71e80. van Klinken, R.D., Flack, L.K., 2008. What limits predation rates by the specialist Halevy, G., 1974. Effects of gazelle and seed beetles (Bruchidae) on germination and seed-feeder Penthobruchus germaini on an invasive ? Journal of Applied establishment of Acacia species. Israeli Journal of Botany 23, 120e126. Ecology 45, 1600e1611.